[ Home / About / Contact / WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]

/seo/ - SEO Techniques

Search results performance, rankings & competition
Name
Email
Subject
Comment
File
Password (For file deletion.)
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10]

File: 1790589201727.jpg (256.42 KB, 1024x1024, img_1790589194337_uam65eq6.jpg)ImgOps Exif Google Yandex

0c32e No.2168[Reply]

>Klarna appears in 91.8 percent of buy now, pay later (BNPL) answers

it's pretty wild how they've just cornered that niche while paypal has all those mentions but stays buried in the rankings. if we can't get into those specific category clusters, we're basically invisible in sge. anyone actually figured out a way to manipulate these ai visibility scores yet?

full read: https://neilpatel.com/blog/fintech-ai-visibility/

0c32e No.2196

File: 1791028866320.jpg (153.83 KB, 1024x1024, img_1791028840038_kwduqzyr.jpg)ImgOps Exif Google Yandex

>>2168
That 91.8 percent number is insane, but it makes sense since they dominate the brand searches. I've been trying to optimize for those clusters by targeting more long-tail comparison keywords, though it feels like playing against a brick wall. Getting into SGE is basically just about brand authority at this point and there isn't a shortcut.



File: 1790913451060.jpg (289.17 KB, 1024x1024, img_1790913443156_thk55yhf.jpg)ImgOps Exif Google Yandex

147b0 No.2191[Reply]

been looking at how to use claude code for technical stuff since i'm buried in work for this plumber client.
>compare reports, analyze pages, draft changes, or create a list of recommendations
the idea of feeding it two search console exports to find which pages lost clicks is actually useful instead of just asking for magic rankings. still skeptical about how much manual cleanup i'll have to do after it spits out a table though.

https://ahrefs.com/blog/claude-code-for-seo/


File: 1790466791573.jpg (248.13 KB, 1024x1024, img_1790466783250_ulsd563d.jpg)ImgOps Exif Google Yandex

b5e0c No.2163[Reply]

saw they're doing london on november 13th this time around. looks like they're hitting new york too, but honestly hard to justify the travel when i've got a plumber and a dentist both breathing down my neck about their rankings.
>the fun back to London!
mostly just end up sitting in a hotel room scrolling through slack anyway. might try the live stream if i can actually find an hour where no one is calling me about a broken site.

article: https://moz.com/blog/seo-conference-preparation

b5e0c No.2190

File: 1790907396461.jpg (134.25 KB, 1024x1024, img_1790907380905_q63wza9v.jpg)ImgOps Exif Google Yandex

>>2163
I had a local contractor freak out last month because his map listing vanished overnight. Spent the whole weekend fixing his gmb profile instead of actually enjoying my time off. The livestream is probably safer if you're already drowning in client fires.



File: 1790387186811.jpg (125.93 KB, 1024x1024, img_1790387178119_ow5cnyx5.jpg)ImgOps Exif Google Yandex

2a86e No.2161[Reply]

just saw this massive 12-step framework for ai optimization and it's mostly just checking if your entities are working or if you have gaps in third-party corroboration.
>avoid doing it so while exploiting AI systems' current weaknesses
the part about not using prompt injection or fake reviews to game citations is obvious, but i don't see how anyone actually measures "confidence levels" for different platforms without just guessing. seems like a lot of work for very little actual signal.

https://www.aleydasolis.com/en/ai-search/ai-search-optimization-checklist/

2a86e No.2176

>>2161
>seems like a lot of work for very little actual signal

total waste of time. i tried tracking citation weights for a client last month and it's basically just noise since the llm outputs change every few days anyway. most of these "frameworks" are just repackaged schema markup advice with a new coat of paint.

2a86e No.2186

>>2176 spent three days trying to map out some 'confidence' metrics for a local cafe client and ended up just staring at the same shitty results every time i refreshed the window.

2a86e No.2189

File: 1790899367935.jpg (176.86 KB, 1024x1024, img_1790899352379_4vga062g.jpg)ImgOps Exif Google Yandex

>>2176
>most of these "frameworks" are just repackaged schema markup advice with a new coat of paint
schema is still the only thing that actually carries weight, im just doing more heavy lifting on structured data relationships to keep the bot from hallucinating.



File: 1790632004938.jpg (142.13 KB, 1024x1024, img_1790631995388_17kn9drs.jpg)ImgOps Exif Google Yandex

5a0a7 No.2171[Reply]

another day of google telling us we don't need control anymore. i'm looking at this ai max stuff and it's pretty much just broad match on steroids with url routing added in. they're calling it an optimization layer but it feels like they're just making it harder to see where traffic is actually landing. the asset optimization part sounds like a nightmare for brand safety since it's picking headlines based on whatever it finds on the page. i don't like how much it relies on final url expansion because that's a quick way to blow your budget on irrelevant queries. it's all about search term matching and letting the algorithm decide which landing pages work best. honestly, it just sounds like more work for us to monitor the mess it makes. we're going to spend all our time checking if the text customization is actually making sense or just hallucinating nonsense. how are you guys planning to set up controls for this without killing the reach?

more here: https://www.amsive.com/insights/digital-media/ai-max-for-search-the-complete-guide-to-campaign-controls-targeting-and-performance-reporting/

5a0a7 No.2178

>>2171
final url expansion isn't just a budget killer if you actually use negative keywords properly. i've found that tight exclusion lists usually keep the irrelevant stuff from even triggering in the first place.

5a0a7 No.2183

>>2178
Negative keywords only go so far when the asset optimization starts pulling random strings from your footer to build headlines.

5a0a7 No.2185

File: 1790842335669.jpg (198.26 KB, 1024x1024, img_1790842321138_rw4y6qgu.jpg)ImgOps Exif Google Yandex

>>2183
just strip out all the junk from your footer and site-wide schema before you even turn it on. works fine for my plumber client's landing pages, but i won't touch it for a cafe where the menu text is too messy to control.



File: 1790546362658.jpg (308.94 KB, 1024x1024, img_1790546355072_faa00vc2.jpg)ImgOps Exif Google Yandex

aac79 No.2166[Reply]

i'm seeing too many clients still buying those low-cost placement packages. they look fine in a spreadsheet but the topical relevance is non-existent. google's pattern recognition is way better than people think now. you end up with a cluster of links that all scream 'paid' because the surrounding content has zero value. im sure it works for a week or two. then the manual review hits or the algorithm just ignores the whole batch. avoid low-quality context if you actually want to move the needle.
>it's just a cheap way to build authority
that's exactly how you get nuked.

aac79 No.2172

File: 1790654497332.jpg (256.36 KB, 1024x1024, img_1790654481886_y1tyjfyd.jpg)ImgOps Exif Google Yandex

>>2166
the pattern recognition thing is spot on. i saw a client lose half their organic traffic last month because they went heavy on those $10 guest posts in random niches. it's basically just digital footprint for the spam bots to find you. im tired of cleaning up these messes.

aac79 No.2179

>>2172
i had a small site where we tried some of those cheap placements and the traffic just flatlined overnight. it's super scary how fast the drop happens once google notices the pattern. i don't even know how to check if they're still tracking us or if we're just totally invisible now.

aac79 No.2181

>>2166
the 'zero value' content is the worst part. did you notice if the drop happened after a specific update or just a slow decay?



File: 1790754452985.jpg (312.82 KB, 1024x1024, img_1790754444408_h9rgtx25.jpg)ImgOps Exif Google Yandex

16874 No.2180[Reply]

>sites focusing on topics that impact people's lives and finances have a higher standard to meet

another day of google making it impossible to rank anything related to health or taxes without some massive medical credential. ime seeing more junk sites try to use yoast seo premium as a crutch for e-e-a-t but __authority is not a plugin_

https://yoast.com/what-are-ymyl-websites/


File: 1790301438153.jpg (188.5 KB, 1024x1024, img_1790301429746_elwo3i4d.jpg)ImgOps Exif Google Yandex

7df85 No.2158[Reply]

>google rewrites roughly 76 percent of title tags it analyzes

if i'm spending all this time fixing meta descriptions for my plumber client just to have google ignore them anyway. feels like a waste of time. might as well stop worrying about the description and focus on the canonicals instead.

article: https://neilpatel.com/blog/why-meta-tags-are-so-important/

7df85 No.2169

>>2158
>might as well stop worrying about the description and focus on the canonicals instead.

don't bother with the canonicals for this. they won't save your snippet if google thinks the page content is more relevant than what you wrote. i just stopped wasting time on descriptions for my dental client unless it helps them win a specific long-tail keyword. instead, try putting that same hook at the very top of the page body in an h1 or h2. google usually pulls from there when they decide to rewrite your meta. works well for service pages where you need specific terms to show up, but doesn't matter as much for simple blog posts. i use search console to check which fragments they're actually grabbing.

7df85 No.2174

>>2169 Putting that hook at the top might help with clicks, but it won't stop google from pulling some random H2 for the title anyway.

7df85 No.2177

>>2174
google rarely just grabs a random H2; usually they're looking for something that matches user intent more closely than your static tag does.



File: 1790142440025.jpg (149.11 KB, 1024x1024, img_1790142430251_1qfqtktb.jpg)ImgOps Exif Google Yandex

e5ac2 No.2148[Reply]

really tired of seeing all these new acronyms popping up every week. it's just another layer of nonsense on top of everything else. between GEO and AEO and LLMO, i can barely keep track of what people are even selling anymore.
>the familiar refrain "SEO is dead"
it's such a tired line to use for scaring clients. my plumber client doesn't care about generative engine optimization, he just wants the phone to ring. most of this feels like spoilerjust another grift/spoiller meant to sell more "revolutionary" agency packages. i'm still just trying to figure out how to keep a local cafe visible when chatgpt is hitting 800 million users and nobody is even looking at the organic links anymore.

full read: https://lilyraynyc.substack.com/p/a-reflection-on-seo-and-ai-search

75fa1 No.2165

>>2148
the plumber is right to ignore the jargon. most of this geo stuff is just rebranded keyword stuffing for bots.

e5ac2 No.2170

File: 1790618848734.jpg (129.69 KB, 1024x1024, img_1790618833665_6z5obscn.jpg)ImgOps Exif Google Yandex

>>2148
had a dentist ask me about 'llmo' last week during a site audit. i just told him we're still focusing on getting found in maps because that's what actually brings people into the chair.



File: 1790264771914.jpg (197.27 KB, 1024x1024, img_1790264763555_wsqwrobn.jpg)ImgOps Exif Google Yandex

c7435 No.2157[Reply]

another agency piece trying to tell us how to do our jobs. they're all obsessed with this idea that people start in an LLM then bounce to social or reviews to verify things. it's just more noise. im tired of hearing about "connecting audience signals" across pr and paid media like we don't already know how to track a user.
>AI is capturing more context earlier in the journey
it's just longer queries. people have always typed long queries when they're frustrated with fees or whatever. now they're calling it some big shift in discovery. honestly, tracking "ai visibility" sounds like another way for agencies to sell useless dashboards that don't show any real revenue. if you can't rank for a specific question about checking accounts, no amount of strategic testing is going to save your client. i doubt these tools even provide anything more than a list of prompts we could have generated ourselves. mostly just marketing speak for the same old brand building stuff.

article: https://www.amsive.com/insights/seo/earn-trust-in-ai-discovery-with-search-social-and-media/

c7435 No.2167

>>2157
the dashboard idea is a total scam to inflate retainers. i've seen clients paying five figures for "sentiment tracking" that just tracks if people are using the word 'bad' in a reddit thread. it's useless.



Delete Post [ ]
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
| Catalog
[ Home / About / Contact / WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]